J . Phys. Chem. 1994, 98, 1398-1406
1398
Bond Strengths of Transition Metal Diatomics: Zr2, YCo, YNi, ZrCo, ZrNi, NbCo, and NbNi
Caleb A. Arrington, Thorsten Blume, and Michael D. Morse'
Department of Chemistry, University of Utah, Salt Lake City, Utah 84112
Mats DoventPl and Ulf Sassenberg
Department of Physics, Stockholm University, Vanadisvagen 9, 1 1 346 Stockholm, Sweden
Received: September 27, 1993; In Final Form: November 28, 1993"
The observation of an abrupt predissociation threshold in an extremely congested electronic spectrum has been
used to measure the bond dissociation energies of the mixed early-late transition metal molecules YCo, YNi,
ZrCo, ZrNi, NbCo, and NbNi. In these systems it is argued that predissociation occurs as soon as the
ground separated atom limit is exceeded, providing values of Di(YCo) = 2.591 f 0.001 eV, Di(YNi) = 2.904
f 0.001 eV, D:(ZrCo) = 3.137 f 0.001 eV, Di(ZrNi) = 2.861 f 0.001 eV, Di(NbCo) = 2.729 f 0.001 eV,
and Di(NbNi) = 2.780 f 0.001 eV. In addition, the bond strength of diatomic zirconium has been measured
as Di(Zr,) = 3.052 f 0.001 eV. A discussion of the chemical bonding in the mixed early-late transition metal
dimers and Zr1 is presented, based on the measured bond energies of these species.
I. Introduction
The chemical bonding between transition metal atoms is an
area of interest to many different branches of chemistry, including
metallurgy, surface science, and metal cluster chemistry. Unraveling the details of the complicated electronic structure in
these metallic species is an ongoing process pursued through both
experimental1.2 and theoreticaP4 avenues. From this research it
is already clear that several distinct factors combine to determine
thestrength and character of the chemical bond between transition
metal atoms. One such factor is the loss of exchange energy
which occurs upon formation of d-electron bonds, thereby reducing
the strength of the chemical bonding. This is illustrated
in the example of Ti2,5 where exchange effects favor the
4sui3d~:3du:3d6:, 3As,, state as a candidate for the ground state,
while bonding considerations favor the 4sui3d?r~3du~,
IZ; state.
In Ti2 the balance between exchange effects and chemical bonding
is so delicate that ab initio theory is incapable of determining
which possibility is the ground state.5 Experimentally it is found
that exchange effects dominate in Tiz, leading to an X 3$,, ground
state.6 In the isovalent Zr2 molecule, however, the larger d-orbitals
favor bond formation, leading to a strong theoretical prediction
of a '2: ground
Thus, to understand the bonding in the
diatomic transition metals the competing effects of exchange and
chemical bonding must be carefully considered.
Another competing factor which plagues the understanding of
chemical bonding in the diatomic transition metals is that often
the strongest chemical bond results from combining excited-state
atoms. For transition metals with ground states of d@, the
strongest chemical bonds generally result when the atoms are
promoted to the excited dn+W atomic limit. This excited
configuration of the atoms is now capable of forming a strong su2
bond, and in the early transition metals strong d-orbital bonding
is possible as well. In systems where d s promotion is required
to reach the dn+W limit, however, it can be difficult to determine
whether the bonding energy gained is sufficient to overcome the
energetic cost of promotion to the excited dn+W configuration.
Work in this group has concentrated on investigating the
electronic structure of diatomic transition metals through spectroscopic methods.6~*-~5
In addition to the spectroscopic determination of the ground-state term symbol, electronic configuration, bond length, and vibrational frequency, another piece of
-
Abstract published in Aduance ACS Abstracts, January 15, 1994.
0022-365419412098-1398%04.50/0
information which is valuable for understanding the nature of
transition metal bonding is the bond dissociation energy, which
represents the sum of all the interactions occurring between the
atoms. In the present investigation we report bond strength
measurements of several mixed early-late transition metal
diatomics, specifically the diatomic combinations of the early 4d
metals yttrium, zirconium, and niobium with the late 3d metals
cobalt and nickel. These metal species are of interest because of
the combination of an electropositive early transition metal with
an electronegative late transition metal, which creates the
possibility of an ionic contribution to the bonding by donation of
s electron density from the early transition metal to the late
transition metal. In addition to this transfer of s-electron density
to the more electronegative late transition metal, there is also the
possibility of significant back-donation of d-electron density to
the early transition metal atom, which is d-electron deficient.
This is the essence of the Brewer-Engel model of transition metal
bonding,l6 which predicts the mixed early-late transition metal
molecules to be more strongly bound than their homonuclear
counterparts. Measurements of the bond strengths of these
molecules will help to clarify the applicability of this model to
the smallest systems.
Previously, the bond dissociation energies of TiV,6 V2,8 TiCo,6
VNi,8 Ni2,9NiPt,loPtZ,llAlNi,l2C0~+,13Ti~+,~~
V2+,14and C03+ l 4
have been measured by the abrupt onset of predissociation in an
extremely congested optical spectrum. To this list we now add
seven new bond strengths measured by this method: Zrz, YCo,
YNi, ZrCo, ZrNi, NbCo, and NbNi. The factors contributing
to the bonding in these species, along with periodic trends among
the growing number of intermetallic metals which have been
investigated, are also discussed.
11. Experimental Section
The intermetallic molecules listed above are created by pulsed
laser vaporization of a metal alloy disk in the throat of a pulsed
supersonic expansion of helium. The ablated metal atoms are
expelled into a 2-mm channel just prior to the peak density of the
helium pulse, are entrained in the helium flow, and undergo
supersonic expansion into vacuum from a 2-mm orifice approximately 3 cm downstream from the point of vaporization.
Although larger clusters are also produced, by carefully controlling
the firing time of the vaporization laser (Nd:YAG second
harmonic, 15 mJ/pulse) and the backing pressure of helium, one
0 1994 American Chemical Society
Bond Strengths of Transition Metal Diatomics
may decrease the number of larger clusters produced to a
minimum. This reduces the possibility of fragmentation of larger
clusters, which can obscure the true spectroscopic processes
occurring in the diatomic metals. In the present study of the
intermetallic metal dimer bond strengths, a resonant two-photon
ionization (R2PI) process is used to investigate the species present
in the molecular beam. The specifics of the R2PI instrument are
more completely described elsewhere.”
Based on previous studies from this group, rotational temperatures of 3-10 K and vibronic temperatures of 100-500 K are
expected from this source. The presence of vibronically hot
molecules in the molecular beam should not affect the interpretation of the data, because these hot molecules possess a minimum
of 50-100 cm-I of vibronic energy and should therefore predissociate at energies at least 50-100 cm-1 below that required for
molecules in the ground state, which in any case constitute the
majority of species in the beam. Our measured predissociation
thresholds are typically defined to f 2 cm-1, which is inconsistent
with any broadening of the threshold which might be associated
with the population of excited vibronic levels. At rotational
temperatures of 3-10 K, the rotational energy of the molecules
is 2-7 cm-I, and on this basis an uncertainty of fO.OO1 eV (f8
cm-1) is assigned to the measured bond energies.
An unusual and powerful aspect of this molecular beam
apparatus is the use of a metal alloy target disk instead of the
more conventional metal rod source. The metal alloy source disks
used in this investigation were produced in an electric arc furnace
capable of reaching a temperature of 3500 OC. The sample is
prepared by placing weighed amounts of the individual lump
metals, approximately 10-15 g total, in the electric arc furnace.
An arc (100 A, 40 V) is then struckabove the metal pieces, which
are enclosed in an inert atmosphere of argon. Complete melting
of the metals is typically achieved when the arc current is increased
to 200 A. After cooling, the alloyed metal nugget is turned over
and remelted. After three or four repetitions, the convection
currents in the molten metal lead to a completely homogeneous
mixed metal sample. The alloy disk is then ground to provide
a flat surface for laser vaporization, which greatlyreduces intensity
fluctuations in the molecular beam as the sample is translated
and rotated to present a fresh surface for laser vaporization.
The mole ratio of the constituent metals in the alloy can exert
a powerful influence on the abundance of the diatomic metals in
the molecular beam. Our experimentally developed rule of thumb
is that the ratio of the bond strengths of the respective dimers
should be inversely related to the elemental concentrations in the
alloy in order to optimize production of the intermetallic dimer.
As an example, the bond strength of Ni2 is 2.068 f 0.010 eV,9
roughly a factor of 3 less than the bond strength of Nbz (5.21 f
0.10 eV).I* Accordingly, a 1:3 alloy of Nb:Ni was prepared to
study the NbNi molecule. By using an alloy that was enriched
in the element forming the weaker bonds, the exothermic
displacement reaction
NbNi
+Nb
-
Nb,
+ Ni
could be partially overcome, allowing a usable concentration of
NbNi molecules to be produced. For the studies of the YNi,
YCo, ZrNi, and ZrCo molecules, alloys were prepared in a 1:l
ratio, while a 1:3 molar ratio alloy of NbCo was prepared for
investigations of this molecule.
The jet-cooled mixed metal cluster beam passes through a
1-cm skimmer, sending a collimated beam into the ionization
region of a time-of-flight mass spectrometer, where a Nd:YAG
pumped dye laser is used to excite the molecules by collinear
propagation down the molecular beam axis. Excited molecules
are then ionized by the output of an excimer laser operating on
KrF (5.00 eV, 248 nm), which crosses the molecular beam a t
right angles and is fired about 10-20 ns after the excitation dye
laser. A mass-selected optical spectrum of ion signal us frequency
The Journal of Physical Chemistry, Vol. 98, No. 5, 1994 1399
is collected by scanning the dye laser. For most of the molecules
investigated, the dye laser was calibrated using well-known
transitions of atomic z i r ~ o n i u m .In
~ ~the case of YNi, however,
the dye laser was calibrated by narrowing the output using an
intracavity etalon (0.04 cm-1) and pressure scanning with Freon12 (CC12F2, DuPont). Theoutput was Ramanshifted using highpressure (500-psi) H2, and the Stokes-shifted radiation was used
to collect an I2 absorption spectrum for comparison to the atlas
of Gerstenkorn and Luc.20 Using the known Raman shift
(4155.163 cm-1)21of the HZQ(l) transition a t 500 psi (34 atm)
allowed the fundamental radiation of the dye laser to be precisely
calibrated. For all of the molecules studied, a generous absolute
error estimate of f8 cm-I (fO.OO1 eV) is assigned to the
dissociation energy, even though the dye laser was calibrated to
an accuracy of better than f 2 cm-I.
111. Results
In previous studies from this group we have reported abrupt
predissociation thresholds in a congested optical spectrum for
several open d-subshell transition metal diatomics and have argued
that these thresholds correspond to the thermochemical bond
strength of the molecule.*-l4 Our contention has been that in
most of the open d-subshell diatomics the density of electronic
states near the dissociation limit is so high that nonadiabatic
effects render the concept of motion on a single Born-Oppenheimer potential energy surface untenable. Because of nonadiabatic couplings between electronic states, in most examples
a sharp onset of predissociation then occurs as soon as the energy
of separated ground-state atoms is exceeded. In several examples
(such as Ni2,9Pt2,11Co2+,13Ti2+,14andV2+ ‘4) the bond strengths
derived from predissociation thresholds are in excellent agreement
with values determined through.other methods, lending credence
to the argument. In the few cases where abrupt thresholds are
observed which differ from the results of other measurements (as
is the case for Cr2+ 22*23),
specific reasons for the disagreement
are usually evident. In the case of Crz+, for example, the lowest
separated atom limits are ~ S S ~ ( ~ S , d5(6S),
~ S ) and the lack of
orbital angular momentum leads to a low density of electronic
states. Further, all of the states which arise from these limits are
Z+states, which cannot be coupled to one another through spinorbit interactions.25 In another set of examples, NiPd and PdPt
failed to display a sharp predissociation threshold despite a high
density of excited electronic states evident in the spectrum.15These
species possess only attractive potential curves arising from
ground-state atoms, however, and the nesting of the resulting
potential curves may have prevented a sharp onset of predissociation. Even when the lowest separated atom limits generate
only attractive curves, however, a sharp predissociation threshold
may still be identified with the bond strength, as was found in
the case of Co2+.13 In the photodissociation spectrum of this
molecule, a sharp predissociation threshold provides a bond
strength (L$,(Co,+) = 2.765 f 0.001 eV) which is in excellent
agreement with the results of a collision-induced dissociation study
= 2.75 f 0.10eV).26
(D;(Co;)
+
These examples have been previously discussed, and criteria
have been developed as a guide to when an abrupt onset of
predissociation may be expected to occur and to correspond to
the thermochemical bond strength.8J3 The most important of
these criteria is that the molecule must possess a large density
of electronic states. To investigate the expected density of
electronic states in these molecules, it is a simple matter to
determine the number of case (c) potential curves arising from
each separated atom limit. In combining two atoms in spinorbit levels with total angular momentum J A and &, (25.4 +
1)(2JB + 1)/2 distinct relativistic adiabatic (Hund’s case (c))
potential curves are obtained if J.4 + JBis half-integer, while (25<
1)(J> 1) distinct curves result if J A + JB is an integer.
+
+
Arrington et al.
1400 The Journal of Physical Chemistry, Vof. 98, No. 5, 1994
3000.
200
Integrated Density of States, N(E)
B
2
m
I-
vl
B
t
O
Yn
lo00
100
2
D
0
0
0
2
m
4000
6000
Eo00
IW
I
I
I
I
I
21800
21900
22000
22100
22200
Energy above ground stale atoms (cm ')
Frequency (cm-I)
Figure 1. This figure displays the number of Hund's case (c) potential
curves arising within an energy E of ground-stateatoms for YNi, ZrNi,
and NbCo, for energies up to 10 000 cm-l. Abrupt predissociation is
expected in these molecules because of the large density of electronic
states, many of which may be expected to dip below the ground separated
atom limit. The remarkable difference in the density of states between
these moleculesunderscores the complexity of establishing periodic trends
in transition metal compounds.
Summing the number of potential curves arising from all separated
atom limits below an energy, E, then results in an integrated
density of electronic states at the separated atom limit, N ( E ) .
Although this function pertains to the separated atoms, it
nevertheless provides a good indication of the expected state
density in the diatomic molecule. The density of electronic states
calculated in this way for the mixed early-late diatomic molecules
is quite large for all of the species studied here. Nevertheless,
the actual number of states varies significantly from molecule to
molecule. For example, 205 Hund's case (c) potential curves
arise within 10 000 cm-1 of ground separated atoms in YNi while
2780 potential curves arise within the same energy range in NbCo.
Figure 1 displays the integrated density of states, N(E), for the
mixed early-late transition metal dimers YNi, ZrNi, and NbCo.
From the results reported below, it appears that the vastly
different number of potential curves in these species makes no
noticeable difference in the sharpness of the predissociation
threshold. Nevertheless, the differences in state density are
apparent in the optical spectra of these molecules. For example,
NbNi, which generates 1828 potential curves within 10 000 cm-l
of ground-state atoms, shows only a continuous absorption
spectrum below the dissociation limit in Figure 2b, while YNi,
which generates only 205 potential curves within 10 000 cm-I of
the ground separated atom limit, displays discrete vibronic bands
in addition to an underlying continuum in Figure 3b. These two
examples show how an increase in the number of potential curves
causes thevibronicstructure to wash out in the spectra of diatomic
transition metal molecules.
A. Zr2. Although diatomic zirconium is the only molecule in
this study not classified as an early-late species, it provides both
the most interesting spectrum and the strongest evidence of
predissociation at the thermodynamic threshold. The spectrom
of Zr2 was originally collected at the same time as the spectrum
of ZrCo, a terrific advantage of a mass-resolved experiment.
Because a 1:l molar alloy of Zr:Co was employed, the molecular
beam was only 50% zirconium in composition, and theopportunity
for the production of large zirconium clusters was significantly
diminished. We have previously found that photofragmentation
of large clusters can create a false ion signal at the mass of the
dimer, spoiling the possibility of recording a true dimer spectrum.
The elimination of large clusters by use of a mixed alloy target
made it possible to record a resonant two-photon ionization
spectrum of Zr2.
The ground state of Zrz has been theoretically determined to
be 5 s ~ l 4 d ? r ~ 4 d ulZ+,
~ , yielding Q = 0; in Hund's case (c)
n o t a t i ~ n . ~Electric
.~
&pole allowed transitions from the ground
-$
100
n
z
a
: I
O
22100
I
I
I
I
22200
22300
22400
22500
22600
Frequency (cm.l)
Figure2. Resonant two-photon ionization spectraof NbCo (a) and NbNi
(b) taken near their respectivedissociation energies. The NbCospectrum,
collected using a dye laser operatingon coumarin460 in conjunction with
KrF excimer radiation, showsa threshold at 22 014 cm-', yielding a bond
strength of D:(NbCo) = 2.729 & 0.001 eV. The NbNi spectrum was
collected using KrF excimer radiation for ionization and coumarin 440
dye laser radiation for excitationand displays a predissociation threshold
at 22 425 cm-', resulting in a bond strength of Di(NbNi) = 2.780 &
0.001 eV.
state can therefore only populate Q = 0: or Q = 1, states.
However, the combination of two ground-state zirconium atoms
(4d25s2, 3F2) can only lead to 4,, 3,, 3,, 2,, 2,(2), 1,(2), 18(2),
0;(2), and O l ( 3 ) Hund's case (c) states of the dimer.24 Notably
absent from this set are the optically accessible :0 states. Therefore, the 0; states which are accessed when the ground 0; state
absorbs a photon cannot predissociate toground-stateatoms while
preserving their good quantum numbers, Q,g/u, and O+/O-. The
0: states that are accessed should therefore persist in the optical
spectrum until a separated atom limit generating 0: states is
reached. This unusual situation is apparent in Figure 4, where
two predissociation thresholds are evident, at 24613 f 2 and
25183 f 2 cm-I, respectively. It is interesting that the predissociation thresholds are quite evident in the spectrum displayed
in Figure 4, despite the fact that this was a one-color experiment.
By this we mean that the states probed in Figure 4 lie a t an energy
which is more than halfway to the ionization limit, so the same
wavelength used to excite the molecule could also be used to
ionize it. Despite the fact that there was no time delay between
the excitation and ionization laser pulses (since they were in fact
the same pulse), the intially excited state nevertheless predissociates so rapidly that it cannot be ionized by the absorption of
a second photon. This proves that the predissociation occurs on
a time scale faster than 5 ns.
The separation between the two predissociation thresholds (570
f 3 cm-1) matches the spin-orbit excitation energy of the
The Journal of Physical Chemistry, Vol. 98, No. 5, 1994 1401
Bond Strengths of Transition Metal Diatomics
20600
20700
20800
20900
21000
21100
I
I
Frequency (cm.')
I
225/-
23100
I
I
23200
23300
I
(b)
23400
23500
-I
23600
Frequency (cm.l)
Figure 3. Predissociation thresholds in YCo (a) and YNi (b) observed
by resonant two-photon ionizationusing dye laser radiationfrom coumarin
480 and stilbene 420 dyes, respectively, and KrF excimer radiation for
ionization. The predissociation threshold in YCo occurs at 20 898 cm-I,
yielding a bond strength of Di(YCo) = 2.591 i 0.001 eV. In YNi the
predissociation threshold occurs at 23 420 cm-I, providing a bond energy
of Di(YNi) = 2.904 f 0.001 eV.
zirconium atom (571 -41 cm-l)19 within experimental error,
showing that the second threshold corresponds to dissociation to
the Zr(4d25s2, 3F2) Zr(4d25s2, 3F3) spin-orbit excited limit.
This separated atom limit does indeed generate 0: potential
substantiating the idea that predissociation does not
occur until a limit generating the appropriate values of the good
quantum numbers R, g/u, and O+/O- is reached.
This situation in Zr2 is very similar to previous observations
in V2,8 where the 3ZS(O:) ground state creates an identical
mechanism for a double predissociation threshold. In the case
of V2, the separation between the two predissociation thresholds
is also in agreement with the energy separation between the ground
and first excited state separated atom limits. Together, these
double threshold examples suggest strongly that the abrupt
predissociation threshold corresponds to the true bond strength.
This allows us to confidently expect that the abrupt predissociation thresholds measured for other open d-subshell species
will also give accurate bond strengths. We also note that if all
of the 0: states excited were immune to predissociation prior to
the first excited atomic limit, the ratio of the signal intensities
above and below the first threshold should reflect the numbers
of 0: and 1, states in this energy region. If this were the case,
a drop in signal of about two-thirds would be expected at the first
threshold, in stark contrast to the larger drop that is observed.
It is obvious that a (rotationally induced) heterogeneous coupling
mechanism allows 0: states to mix with dissociative 1, states in
this energy region. Another indication of this phenomenon was
our initial failure to detect a second threshold under expansion
+
conditions which populated higher energy rotational states of
Zr2. Only after more extensive rotational cooling was achieved
by using a smaller expansion orifice were the dissociation rates
of the 0: states reduced sufficiently for the second threshold to
be observed. This is in agreement with a heterogeneous coupling
mechanism, where the coupling matrix elements are proportional
to dW.25
Converting the predissociation threshold at 24613 f 2 cm-I,
one obtains a bond strength of D:(Zr2) = 3.052 f 0.001 eV. To
our knowledge, this is the first experimental measurement of this
quantity. Several theoretical calculations and semiempirical
estimates have been made, however. The estimate of D:(Zr2) =
3.22 f 0.2 eV by Miedema and Gir1gerich2~using an extrapolation
from bulk metal properties proves to be the most accurate. Ab
initio quantum chemical treatments of the bonding in Zr2 have
underestimated the bond strength, with Balasubramanian and
Ravimohan reporting a value of D:(Zr2) = 2.45 eV calculated
at the state-averaged complete active space self-consistent field/
multireference singles + doubles configuration interaction (SACASSCF-MRSDCI) level7 and Bauschlicher et al. reporting a
value of D:(Zr2) = 2.21 eV using an averaged coupled-pair
functional treatment.5 Underestimation of dissociation energies
is a common difficulty in calculations of systems which require
an accurate description of d-d bonding, primarily because electron
correlation is more accurately described in the separated atoms
than in the molecule. Improvements in the calculation by
Balasubramanian and Ravimohan including higher-order correlations and extended basis sets have resulted in a final bond
strength prediction of 2.7 eV 5 D:(Zr2) 5 3.0 eV.7 This revisedvalue lies just at the limit of our experimental measurement,
D:(Zr2) = 3.052 f 0.001 eV.
B. YNi and YCo. The ground state of YNi is known to be X
2 2 by matrix isolation ESR spectroscopy.28 From the 8 = 112
ground state, R = 112 and R = 312 excited states are optically
accessible under electric dipole selection rules. The lowest
separated atom limit for YNi, corresponding to Y(4d15s2,2D3/2)
+ Ni(3d84s2,3F4), generates R = 112 and R = 312 states (along
with many others), so the excited states reached upon optical
excitation can fall apart to ground-state atoms while preserving
the good quantum number, 8. Spectroscopically, this is shown
in Figure 3b, where a sharp predissociation threshold is observed
at 23420 f 2 cm-l. Converting to electronvolts and using a more
generous error limit then provide D:(YNi) = 2.904 f 0.001 eV.
This mixed metal bond strength is much greater than that of
either of the homonuclear dimers, Y2 (1.62 f 0.22 eV)Z9 or Ni2
(2.068 f 0.010eV),9inagreement with theBrewer-Engel model.l6
A comprehensive theoretical investigation of the YNi molecule
has been performed by Faegri and Bauschlicher.30 Their study
suggests that the X 22+ground state is best correlated to a
Y(4d15s2) Ni(3d94s1) asymptote, a possibility discussed in
section IV below. As was mentioned in connection with Zrzabove,
such theoretical calculations usually underestimate molecular
bond strengths because the correlation energy of the separated
atoms is more accurately calculated than that of the molecule.
This is also the case in YNi, where the best bond strength estimate
is obtained a t the CASSCF-MRCI level including relativistic
corrections as D:(YNi) = 2.01 eV.30 This severe underestimate
of the YNi bond strength highlights the difficulty in estimating
bond strengths of diatomic transition metals by ab initio methods.
An abrupt predissociation threshold is also found in the resonant
two-photon ionization spectrum of YCo at 20898 f 2 cm-1,
as displayed in Figure 3a. This converts to a bond strength of
D:(YCo) = 2.591 f 0,001 eV. Again this mixed metal diatomic
has a much greater bond strength than either of its homonuclear
counterparts, with D:(Y2) = 1.62 f 0.22 eV29 and D;(Co2) =
0.95 f 0.26 eV.31 In addition, the resonant two-photon ionization
spectrum of YCo shows strong intensity right at the threshold,
+
Arrington et al.
1402 The Journal of Physical Chemistry, Vol. 98, No. 5, 1994
260 -
I
I
I
I
I
I
I
8
’
-
-
-
130-
v,
Dissociation of Zr, to
0 -
I
I
I
I
I
240
4
I
Dissociation of Zr, to
Zr(‘FJ + Zr(’F3
-
4
I
1
I
I
I
I
I
I
I
I
I
I
l
160
-
1
D:(ZrCo) = 3.137 f 0.001 eV
0
I
0
t
24800
24900
25000
25100
25200
25300
25400
25500
25W
25700
25800
Frequency (cm-I)
Figure 5. Predissociation threshold of ZrCo, observed by resonant two-photon ionization spectroscopy using exalite 398 and exalite 389 dye laser
radiation in coniunction with KrF excimer laser radiation for Dhotoionization. The observed predissociation threshold at 25 298 cm-I corresponds to
a bond strengthof D:(ZrCo) = 3.137 f 0.001 eV.
which decreases to a more moderate signal at lower frequencies.
The reason for such behavior is not at all understood; similar
spectral rises as one approaches the dissociation threshold are
also observed in TiCo* and VNi? however.
C. ZrCo and ZrNi. The predissociation threshold of ZrCo at
25298 f 2 cm-1, displayed in Figure 5, results in the largest bond
strength measured in this study, D:(ZrCo) = 3.137 f 0.001 eV.
The large bond strength of ZrCo may indicate that of the six
mixed early-late transition metal diatomics studied ZrCo is the
candidate best suited for application of the Brewer-Engel bonding
model.16 In this model the d-electron density of cobalt overlaps
significantly with the zirconium d-orbitals, leading to strong
multiple d-bonds.
Although ZrNi is not bound as stronglyas ZrCo, it nevertheless
possess a significant bond strength of Dg(ZrNi) = 2.861 f 0.001
eV, as measured from the abrupt predissociation threshold shown
in Figure 6 at 23072 f 2 cm-I. A reverse correlation between
bond strength and promotion energy is observed in ZrCo and
ZrNi. Although cobalt must expend 0.482 eV more in promotion
energy than nickel,l9 the measured bond strength of ZrCo is
nevertheless greater than that of ZrNi. This indicates a significant
d-orbital bonding contribution in these molecules; otherwise, one
would expect the difference in bond strengths to correlatedirectly
with the difference in promotion energies.
D. NbNi and NbCo. Diatomic NbNi and NbCo show sharp
predissociation thresholds in their resonant two-photon ionization
spectra a t 22425 f 2 and 22014 & 2 cm-1, respectively, as shown
in Figure 2. These correspond to bond strengths of Dg
(NbNi) = 2.780 f 0.001 eV and D:(NbCo) = 2.729 f 0.001 eV.
The close similarity in bond strength between these two compounds
200
-
2.891 f 0.001 eV
100
0
22800
22900
23000
23100
23200
Frequency (cm-1)
Figure 6. Resonant two-photon ionization spectrum of ZrNi, exhibiting
a predissociation threshold at 23 072 cm-I. The spectrum was recorded
using coumarin 440 dye laser radiation for excitation along with KrF
excimer radiation for ionization. The measured predissociation threshold
corresponds to a bond strength of D:(ZrNi) = 2.891 i 0.001 eV.
does not reflect the 0.482-eV difference in promotion energy
between nickel and cobalt,Ig once again indicating the presence
of other factors which contribute to the bonding.
IV. Discussion
The experimental measurement of the bond strengths of seven
open d-subshell diatomic transition metals (Zrz, YCo, YNi, ZrCo,
ZrNi, NbCo, and NbNi) adds considerably to the increasing
The Journal of Physical Chemistry, Vol. 98, No. 5, 1994 1403
Bond Strengths of Transition Metal Diatomics
TABLE 1: Bond Strengths of Selected Transition Metal Diatomics
molecule
(AB)
Ti2
TiCo
VNi
Zr2
YCO
YNi
ZrCo
ZrNi
NbCo
NbNi
no. of
d+s
electrons
8
13
15
8
12
13
13
14
14
15
DO,
D:+P(B)'
(eV)
(ev)
1.23 f 0.17c 2.34 f 0.17
3.049
2.40lC
2.10P
2.266
3.05Zf
4.048
3.239
2.591f
3.070
2.904,
3.785
3.137f
3.027
2.861f
3.377
2.729
2.78of
2.946
D:
+P(B)"
P(A) +
(ev)
3.46 f 0.17
4.162
2.902
5.044
4.688
4.519
4.781
4.023
3.901
3.470
(r)nd
4
(coinage analog)
atom
(ev)
A
2.03 k 0.02 (Ct12)~ 0.793
2.03 f 0.02 ( C U ~ ) ~0.793
2.03 f 0.02 ( C U ~ ) ~0.720
1.64 f 0.04 (Ag2)g
1.156
1.74 f 0.10 (CuAg)h 1.329
1.74 f 0.10 (CuAg)h 1.329
1.74 f 0.10 (CuAg)h 1.156
1.74 f 0.10 (CuAg)h 1.156
1.74 f 0.10 (CuAg)h 1.127
1.74 f 0.10 (CuAg)h 1.127
(W
atom
B
0.793
0.544
0.514
1.156
0.544
0.514
0.544
0.514
0.544
0.514
ground configuration and term
ref
4su23dr:3du'3d6', 3Ar
4su~3dd3du~3d6%4su*',
*Z+
4~&3dd3du23d6~3d6*24~~*',
'25su24dx:4du2, IZ;
s a d d d u 2 d 6 k r * ' , 3Al
s ~ * d d d u ~ d 6 ~ s2Zt
u*~,
s~2dddu2d6~su*l,
2Zt
s~2dddu2d6~d6*'su*l,
3Ar
s ~ ~ d d d o ~ d 6 ~ d 6 * ' s'
u4
*',
s ~ ~ d d d u ~ d 6 ~ d 6 *42~so*~,
6
36
38
597
i
28
i
i
i
38
0 Intrinsic bond energies are defined in the text as the measured bond strength plus the promotion energy required to prepare the atoms for bonding,
where the promotion energy of atom B, P(B), is defined by performing a weighted average of the spin-orbit levels of the high- and low-spin couplings
of the SI electron to the lowest term of the dncore. This gives the energies of these terms in the absence of spin-orbit coupling. These are then averaged
to obtain the energy of the hypothetical spin-decoupled sl-dn state above the ground state, which is defined as P(B). In the column headed 4 P(B),
only the late transition metal atom is assumed to promote, while in the column headed D: + P(A) + P(B) it is assumed that both atoms are promoted.
Atomic energy levels are taken from ref 19. From ref 33. From ref 2. From ref 39. From ref 8. /This work. 8 From ref 40. From ref 35.
'Suggested to be the ground state in the present investigation.
+
pool of information available about the chemical bonding in these
complex transition metal systems. A summary of the measured
bond strengths and other pertinent information about these species
is provided in Table 1.
A problem in understanding the chemical bonding in the
transition metal diatomic molecules is the question of which
separated atom limit leads to the ground state of the molecule.
As mentioned in section I, the strongest chemical bonds generally
result from the interaction of a dn+'sl atom with a dm+W atom,
which leads to a strong s-based a2 bond. Although this can also
arise from the interaction of a d"s2 atom with a dm+2s0atom, this
separated asymptote nearly always lies considerably higher in
energy than the dn+lsl dm+W limit in the neutral transition
metal diatomics. Thus, in most cases the ground state of the
separated atoms cannot correlate to thestrongly bound saZd*ldm+l
molecular configuration, and an amount of energy must be
expended to excite the atoms to an appropriate separated atom
limit in preparation for bonding. As a consequence, the ground
state of the molecule often may depend on a delicate balance
between the energy gain due to chemical bonding and the energetic
cost of preparing the atoms for bonding (Le., their promotion
energy).
In computing the promotion energy, it is useful to remember
that when one forms an sdchemical bond between a dn+W atom
and a dm+W atom, one does not gain the full strength of the sa2
bond because of the loss of the d n + W and dm+l-sl exchange
energy in the high spin coupled dn+W and dm+ls' atomic states.
Thus, one should consider the promotion energy to be the amount
of energy required to excite the atom to a hypothetical state lying
halfway between the high- and low-spin couplings of thed electron
to the dn+l or dm+l core. This approach has been used to calculate
the promotion energies for transition metal atomiccations bonding
to a hydrogen atom, and a clear correlation of bond strength with
promotion energy has been f0und.3~We note that this represents
a refinement in the definition of promotion energy as compared
to our previous use of the term.*
Using the promotion energy defined in this way, we have also
calculated intrinsic bond energies for the molecules reported in
this investigation. These intrinsic bond strengths are defined as the bond strengths that would be expected if an
sa2didr+'sae' AB molecule dissociated diabatically to the dns2
(atom A) the hypothetical spin-decoupled dm+W (atom B)
separated atom asymptote. Intrinsic bond strengths defined in
this way are listed in column four of Table 1 as Di + P(B).
Alternatively, the ground state may be sa'di+'dr+', which
diabatically correlates to the doubly promoted separated atom
limit of dn+W (atom A) + dm+W (atom B), in which case both
+
+
atoms must be promoted to form the ground state of the molecule.
If this is the case, the intrinsic bond strength is given by Di
P(A) + P(B), listed in column five of Table 1, where P(A) and
P(B) represent the promotion energies required to prepare the
atoms in the spin-decoupled dn+Wand dm+Wstates, respectively.
In all cases listed in Table 1, the early transition metal atom (Ti,
V, Y, Zr, or Nb) is designated as atom A and the late transition
metal atom (Co or Ni) is designated as atom B. With the single
exception of NbCo, the third alternative of diabatic correlation
of an sa2di+'d:sa*' state to the spin-decoupled d"+'sl (atom A)
dmsz (atom B) separated atom limit need not be considered,
since this limit lies above the dns2(atom A) dm+W (atom B)
asymptote, and therefore the lower asymptote will generate the
favorable S U ~ S Uconditions
*~
for bonding.
From the preceding commentary, it is clear that the separated
atom parentage of the ground state of the diatomic molecule will
be an important consideration. Unfortunately, in only a few of
the examples considered here is there any evidence, either
experimental or theoretical, that speaks to this point. This question
will be addressed further in the discussions of the individual
molecules below.
A. Zrz. As mentioned above, the homonuclear diatomic
molecule, Ti2, is now experimentally known to possess a
4su;3d?r:3da:3dfi:,
3Ag,r ground state,6 where 4sa-3da hybridization is thought to stabilize the " 4 s ~ ~orbital
"
and destabilize
the "3dag" orbital, resulting in an unexpected orbital filling order.
This ground state correlates to the doubly promoted 3d34s1, 5F
3d34s1, 5F separated atom limit, which greatly decreases the
measured bond strength of the molecule due to the high promotion
energy associated with preparation of the spin-decoupled 3d3('F)4s' state of the titanium atom (1.113 eV per atom) and the
relatively poor 3d bonding that results because of the small size
of the 3d orbitals.
In the isovalent Zrz molecule, on the other hand, the orbital
radii of the 4d and 5s orbitals are more similar ((ru)/(rSS) =
0.549 in Zr 4d25s2,3F)33than are the radii of the analogous 3d
and 4s orbitals in titanium ((r3d)/(rds) = 0.401 in Ti 3d24s2,
3F),33 making d-orbital bonding more favorable in diatomic
zirconium than in diatomic titanium. In addition, the promotion
energy required to prepare a zirconium atom in a spin-decoupled
4d3(4F)-5sl state is reduced slightly from that of titanium, to
0.996 eV per atom. On these grounds we may confidently expect
that the ground state of Zr2 derives from two spin-decoupled
4d3(4F)-5sl atoms. Indeed, a b initio calculations are now
in general agreement t h a t t h e ground s t a t e o f Zr2 is
5sa:4dx:4daj,
This conclusion is also in agreement with
the observation of a double predissociation threshold, as discussed
+
+
+
+
12i.5"
1404 The Journal of Physical Chemistry, Vol. 98, No. 5, 1994
above. Thus, the intrinsic bond strength of Zr2 is 5.044 eV, in
comparison to an intrinsic bond strength of only 3.46 f 0.17 eV
for Ti2 (based on the value D:(Ti2) = 1.23 f 0.17 eV).2
Undoubtedly the enhanced intrinsic bond strength of Zr2 as
compared to Ti2 results from the improved overlap between the
d orbitals in zirconium as compared to titanium. This improved
d-electron bonding also favors spin-pairing of the 4d electrons in
the 4da, orbital, leading to the IZ; state calculated to be the
ground state of Zr2.5v7 In the remaining molecules investigated
in the present study, the compact nature of the 3d orbitals will
continue to play an important role in limiting the extent of d
orbital contributions to the bonding, particularly as one moves
to the right in the 3d transition metal series.
B. YNi and YCo. Van Zee and Weltner have determined the
ground electronic state of matrix-isolated YNi to be 2Zby electron
spin resonance (ESR) spectroscopy.28 Their analysis of the
hyperfine coupling associated with the Z 112 SgY nucleus in
YNi shows the unpaired u electron to have 30% 5sy character
and 80% 4dy character, a result that does not conserve probability.
This apparently erroneous result occurs because the analysis
cannot determine the amounts of 4dy, 5sy, and 5py character
independently, since only two independent parameters (the
isotropic and dipolar components of the s9Y hyperfine coupling
tensor) are measured. The 5py hyperfine interaction is larger
than the 4dy hyperfine interaction, so it is likely that theunpaired
u electron occupies an orbital with substantial 5py character,
contributing strongly to the dipolar component of the hyperfine
interaction tensor. Thus the experimental measurements support
an electronic structure of s ~ ~ d # d u ~ d 6 ~ s2Z+,
u * ~where
,
considerablemixingoftheatomicorbitalsoccurs within the aframework,
making the identification of orbitals as “sa” or “dun tenuous at
best. Indeed, throughout the entire series of early-late transition
metal diatomics investigated here, it is likely that significant sadu hybridization occurs, resulting from the close proximity of the
dn+lsl and dGatomic states. As a result, the nominal du bonding
orbital is destabilized (and the nominal sa bonding orbital is
stabilized), leading to an unusual orbital filling order placing the
du above the d r bonding orbital (as is observed in Ti25,6 and
TiV,8*34which have ground states of 4su~3dr~3dui3d6i,
3Ag,r,
and 4 s ~ ~ 3 d # 3 d d 3 d 642-,
~ , respectively). Likewise, the nominal
sa’ orbital is stabilized (and the nominal du* orbital is destabilized), probably making configurations of su2d#du2d6nd6*msu*
the rule rather than the exception in the molecules investigated
here.
This basic concept of the electronic structure of the ground
state of YNi also finds confirmation in an ab initio investigation
by Faegri and B a u s c h l i ~ h e r .Using
~ ~ a complete active space
self-consistent field (CASSCF) method todetermine thegeometry
of the 2Z+ ground state, these investigators then used a multireference configuration interaction (MRCI) method to more
accurately treat the electron correlation problem. Using the
natural orbitals of the MRCI treatment, valence orbital populationsof 1.84,1.15,0.96,and8.96werefoundforthe(5s+ 5p)y,
(4s + 4 p ) ~ i4dy,
, and 3dNi orbitals, respectively. Considering the
admixture of 5py and 4pNi as providing a means for the more
diffuse 5sy and 4sNi orbitals to polarize, it makes sense to consider
the ground state of the YNi molecule as deriving primarily from
an su2dn4du2d64su*l molecular configuration. Clearly, the
excessive energetic cost of promoting the yttrium atom to the
4d25s1 configuration (1.449 eV) is prohibitive in the case of its
interaction with nickel. On the basis of this ground molecular
configuration, theintrinsic bond strength, which would beobtained
as the actual bond energy if the ground separated atom limit
consisted of Y (4d15s2, 2D) + Ni (spin-decoupled 3d9(2D)-4s1),
is obtained from column four of Table 1 as 3.070 eV. Both this
and the measured bond energy of 2.904 eV are considerably greater
than the bond energy of the coinage metal analogue, CuAg
(D:(CuAg) = 1.74 f 0.10 eV),35 despite the presence of an
Arrington et al.
antibonding electron in the su* orbital. This suggests that
d-orbital contributions; ionic effects due to the combination of
an early, electropositive transition metal with a late, electronegative transition metal; or both combine to substantially increase
the bond strength of YNi beyond what would be expected from
the su framework alone.
Given the excessive promotion energy required to prepare
yttrium in its 5s14d2 configuration, it is likely that the ground
state of YCo corresponds to the removal of a weakly bonding d6 electron from YNi, giving a ground electronic state of
sa2d#du2d63su*l, 3Ai for YCo. Thus, although the bond strength
of YCo (2.591 eV) is considerably reduced from that of YNi
(2.904 eV), the intrinsic bond strengths, calculated assuming
promotion of the cobalt and nickel atoms to the spin-decoupled
3d8(3F)-4s1 and 3d9(2D)-4s1states, respectively, are more nearly
equal, taking values of 3.239 and 3.070 eV for YCo and YNi,
respectively. The remaining difference in bond strengths runs
counter to what one might expect, given that one additional
(weakly) bonding d6 electron is present in the YNi moleculethan
in YCo. However, when one remembers that the 3d orbitals of
nickel are more highly contracted than those of cobalt, and are
hence less accessible for chemical bonding, the measured bond
strengths become more understandable. In addition, if the ground
state of YCo is indeed s ~ ~ d # d u ~ d 6 ~ s u3Ai,
* l ,then an exchange
stabilization between the su* electron and the d6 hole will
contribute to a further stabilization of the YCo bond.
C. ZrCo and ZrNi. The intermetallic dimer, ZrCo, which is
isoelectronic with YNi, has the largest bond strength measured
in the present study, with Di(ZrCo) = 3.137 f 0.001 eV. The
chemical bonding is likely to be quite similar in YNi and ZrCo,
since the isovalent ScNi and TiCo molecules have been shown
through matrix isolation ESR methods to bond with similar
ground-state configurations of 4sa23d#3du23d644su*l, 22+.36
Accordingly, we expect an s u 2 d ~ d ~ ~ d 6 ~22+
s u *ground
~ , state
for ZrCo, in analogy to that found for other 13-valence-electron
early-late transition metal dimers such as YNi,28YPd,28ScNi,2*.36
ScPd,28 and TiCo.36
Despite the similar ground-state Configurations expected for
YNi and ZrCo, the bond strength of ZrCo (Di(ZrCo) = 3.137
f 0.001 eV) is significantly greater than that obtained for YNi
(Di(YNi) = 2.904 f 0.001 eV). Moreover, this difference in
bond strength is accentuated when the intrinsic bond strength is
calculated by adding the promotion energy of cobalt and nickel
to the values of Di(ZrCo) and Di(YNi), giving 3.785 f 0.001
eV and 3.070 f 0.001 eV for ZrCo and YNi, respectively. This
may be explained by the differential contraction of the 3d us 4s
orbitals in the 3d transition metal series as one traverses the
series from left to right. This differential contraction causes the
3d orbitals to become chemically rather inaccessible in nickel, a
result which is the basis for a recent success in describing the
electronic structure of NiCu and Ni2 by a ligand field method.37
Thus, the d-orbital contributions to the chemical bonding are
probably significantly reduced in YNi as compared to ZrCo,
accounting for the weaker bond in the former molecule. A similar
effect has been previously noted in comparing the bond strengths
of TiCo (D:(TiCo) = 2.401 eV) and VNi (Di(VNi) = 2.100
eV).s Although the electronic configuration of VNi
(4sa23d#3du23d643d6*24su*~, 42-) differs from that of TiCo
(4s~23dn43da23d6~4su*~,
2Z+) by the addition of two nominally
antibonding 6* electrons, this is insufficient to explain the
tremendous difference in intrinsic bond strengths of TiCo (3.049
eV) as compared to VNi (2.266 eV). Again, the highly contracted
nature of the 3d orbitals in nickel greatly reduces their ability to
participate in d orbital bonding. This trend is also evident in the
molecules investigated in the present study, where the intrinsic
bond strengths for the nickel compounds YNi, ZrNi, and NbNi
are all significantly less than those of YCo, ZrCo, and NbCo (see
Table l ) , regardless of whether one assumes that all of these
Bond Strengths of Transition Metal Diatomics
The Journal of Physical Chemistry, Vol. 98, No. 5, 1994 1405
molecules derive from promotion of just the late transition metal
contribution of d bonding in the NbNi species. These effects are
atom or from promotion of both atoms.
almost exactly balanced by the higher promotion energy of cobalt
as compared to nickel, giving NbCo and NbNi bond dissociation
The intermetallic dimer, ZrNi, with D,O(ZrNi) = 2.861 f
energies which are identical to within 0.05 eV.
0.001 eV, is a 14-valence-electronsystem, with one more electron
This work demonstrates that a number of distinct factors are
than either YNi or ZrCo. Assuming that the high promotion
important in determining the strength of the chemical bond in
energy of zirconium (0.996 eV) forces the molecule to adopt an
the intermetallic early-late transition metal diatomics. There is
SU~SU
configuration,
*~
the most plausible candidate for the ground
clear evidence that the late metal atom (cobalt or nickel) is
state is s~2d+da2db4d6*~su*~,
3Ar,deriving from the Zr (4d25s2,
promoted to the d"s1 configuration in all of the compounds that
3F) Ni (3d94sl,3D) separated atom limit. Alternatively, if the
have been investigated, and the different promotion energies of
zirconium atom can be promoted, the s ~ ~ d a ~ d u ~ d 632~ dstate
b*~,
cobalt us nickel are reflected in the measured bond dissociation
deriving from the Zr (4d35sl, SF) Ni (3d94s1,3D) separated
energies. In addition, the reduced size of the 3d orbitals in nickel
atom limit would be the most likely candidate for the ground
causes the d-orbital contributions to the bond strengths of the
state. Given the very similar bond strengths of YNi (08
nickel compounds to be reduced compared to those of the cobalt
(YNi) = 2.904 eV) and ZrNi (D:(ZrNi) = 2.861 eV) and the
compounds. Thus, in moving from YNi to ZrNi and on to NbNi,
large (and different) promotion energies of yttrium (1.449 eV)
one sequentially adds d6* electrons to the system, but this only
and zirconium (0.996 eV), it seems likely that in neither molecule
decreases the intrinsic bond strength by 0.043 and 0.081 eV in
is the early transition metal atom promoted. This argues in favor
moving from YNi to ZrNi and from ZrNi to NbNi, respectively.
of an su2dddu2db4db*lsu*~,3 4 ground electronic state for ZrNi.
In contrast to this minor dependence of the intrinsic bond strength
The slight decrease in bond strength in going from YNi to ZrNi
on the number of d electrons in the nickel systems, the cobalt
then corresponds to the addition of a weakly antibonding d6*
molecules show a rather substantial dependence. The ZrCo
electron in the latter molecule.
molecule, for example, shows by far the greatest intrinsic bond
D. NbNi and NbCo. Niobium differs from yttrium and
strength of the mixed early-late transition metal diatomics
zirconium in having a 4d45s1,6Dground electronicstate. Although
investigated in this series, and it is also the molecule which is
this would lead one to expect that there would be no promotion
thought to have an electron configuration in which all of the d
required to prepare the niobium atom for bonding, the hypothetical
bonding orbitals are filled. Removing a bbnding db electron from
spin-decoupled4d4(SD)-5s1 state nevertheless lies 0.524 eV above
ZrCo
to form YCo reduces the intrinsic bond strength by 0.546
ground-state atoms. Thus the promotion energy associated with
eV, while adding an antibonding d6* electron to ZrCo to form
Y, Zr, and Nb falls as one moves across the series, giving 1.449,
NbCo decreases the intrinsic bond strength by 0.408 eV. Thus,
0.996, and 0.524 eV, respectively. The smaller promotion energy
the d6 and db* orbitals are essentially nonbonding in the nickel
required to prepare niobium in the spin-decoupled 4d4(5D)-5s1
compounds but contribute substantially to the bond strength in
state opens up the possibility of su2 ground states for NbNi and
the cobalt compounds. Similar considerations apply in the
NbCo. This possibility was dismissed in the cases of YCo, YNi,
intermetallic 3d compounds TiCo and VNi, where the intrinsic
ZrCo, and ZrNi but is much more plausible in the case of the
bond
strength of TiCo exceeds that of VNi by 0.783 eV,B a fact
niobium compounds.
which cannot be explained by the mere presence of two additional
Both NbCo (D:(NbCo) = 2.729 eV) and NbNi (08
antibonding db* electrons in VNi. In these compounds it is again
(NbNi) = 2.780 eV) have bond strengths which are somewhat
evident that the multiple d-electron bonds in TiCo are much
reduced from their neighbors, ZrCo (Di(ZrCo) = 3.137 eV) and
stronger than those in VNi, primarily due to the compact nature
ZrNi (D:(ZrNi) = 2.861 eV). This would be consistent with the
of the 3d orbitals in nickel (and vanadium).
addition of a weakly antibonding db* electron to the postulated
ground electronic states of ZrCo and ZrNi, suggesting ground
V. Conclusion
states of s~Zd?r4duZdb~db*~su*~,
3Ar and s ~ ~ d + d u ~ d 6 ~ d 6 * 2 s u * ~ ,
The bond energies of seven diatomic transition metal molecules
42-for NbCo and NbNi, respectively. Although one might think
have been measured by the observation of an abrupt predissoan suZd*4du2db4db*2, 3 2 - and an s~2d+du2db~db*~,
zAi ground
ciation threshold in an extremely congested optical spectrum,
state would be conceivable for NbCo and NbNi, respectively, the
resulting in measured values of Di(YCo) = 2.591 f 0.001 eV,
ESR spectrum of NbNi conclusively demonstrates that the
molecule exists as a 42molecule when isolated in a neon matrix
Di(YNi) = 2.904 f 0.001 eV, D;(ZrCo) = 3.137 f 0.001 eV,
at 4 K.38 Unfortunately, no hyperfine splitting due to the magnetic
Di(ZrNi) = 2.861 f 0.001 eV, D:(NbCo) = 2.729 f 0.001 eV,
61Ni ( I = 3/2) nucleus was observed,so no information concerning
Di(NbNi) = 2.780 f 0.001 eV, and D;(Zr,) = 3.052 f 0.001
the unpaired spin density on the nickel atom could be derived.
eV. The first six of these species may be classified as early-late
However, a large hyperfine splitting due to the niobium 93Nb( I
transition metal diatomics, in which bonding may be partially
= 9/2) nucleus was observed, and the isotropic component of the
described by a Brewer-Engel model. The large bond energies
A tensor was used to derive that the u orbital in the 6 2 ~ 1 ,42obtained in these cases argue in favor of a mechanism in which
ground state is 28% 5SNb in character. Presumably a large portion
s-electron density is donated to the late transition metal atom,
of the remaining character of this u orbital is associated with the
with a concomitant back-donation of d-electron density to the
4sNi orbital, with the remainder associated with the SPNb, 4du~b,
d-electron deficient early transition metal atom.
4 p ~ i ,and possibly the 3dU~iorbitals. The observation of an
The homonuclear Zr2 molecule displays an interesting double
s~2d*~dZdb~d6*2su*~,
4Z1-ground state for NbNi argues strongly
predissociation threshold corresponding to dissociation to both
for an s~2d*~dazdb~db*~su*~,
3Ar ground state in NbCo. Acthe ground 4d25s2,3F2 + 4d25sz, 3F2 and the spin-orbit excited
cordingly, we compute their intrinsic bond strengths as 08
4d25s2, 3F3 + 4d25s2, 3F2 separated atom limits. The exact
(NbNi) + P(Ni) (giving 2.946 eV for NbNi) and D,O(NbCo) +
correspondence of the separation between these thresholds (570
P(Co) (giving 3.377 eV for NbCo), although significant du-sa
f 3 cm-1) withthespin-orbitintervalinatomiczirconium(571.41
mixing would imply that both separated atom asymptotes
cm-I) indicates that predissociation occurs abruptly as soon as
contribute substantially to the bonding.
the excited molecule has sufficient energy to dissociate to a limit
which generates the appropriate values of the good quantum
The decrease of 0.431 eV in intrinsic bond strength in moving
numbers R,g/u, and O+/O-. The previous observation of a similar
from NbCo to NbNi reflects the addition of a second electron
to the weakly antibonding 6* orbital. It seems more likely,
double threshold in Vz provides strong support for assignment of
however, that the major effect in moving from cobalt to nickel
the predissociation thresholds observed in the present work to
is the decrease in the size of the 3d orbitals, which reduces the
actual bond dissociation energies. In addition, Zrz has by far the
+
+
1406 The Journal of Physical Chemistry, Vol. 98, No. 5, 1994
greatest intrinsic bond strength of any of the species presently
investigated. This is thought to result from the perfect size match
of the 4d-orbitals on the two metal centers, combined with their
significant spatial extent, which allows 4d bonds to form without
difficulty.
The chemical bonding in the early-late transition metal
diatomics is discussed in terms of three major factors, which
combine to determine the strength of the chemical bond. First
among these is the promotion energy, which is required to prepare
the metal atoms for chemical bonding. In the examplesconsidered
here it appears that in all cases the cobalt or nickel atom is
promoted to a dnsl state prior to bonding. A second consideration
of considerable importance is the extent of d-orbital overlap
between the early and late transition metals. The compact nature
of the 3d-orbitals on nickel causes the d-orbital contribution to
the chemical bonding to be reduced in all of the nickel compounds
studied, as compared to the corresponding cobalt compounds.
Finally, the number of d-electrons available for bonding determines the occupation of the d-based bonding and antibonding
orbitals, and in thecobalt-containing molecules either the addition
of a d6* electron to ZrCo to form NbCo or the removal of a d6
electron from ZrCo to form YCo leads to a significant reduction
in bond strength.
Acknowledgment. We gratefully acknowledge research support
from U.S.National Science Foundation under Grant No. CHE9215193. Acknowledgment is also made to the donors of the
Petroleum Research Fund, administered by the American
Chemical Society, for partial support of this research. Financial
support from the Knut and Alice Wallenberg Foundation and
the Swedish National Research Council (Contract No. F.FU
3867-309) is also acknowledged.
References and Notes
(1) Weltner, W., Jr.; Van Zee, R. J. Ann. Rev. Phys. Chem. 1984, 35,
291.
(2) Morse, M. D. Chem. Rev. 1986, 86, 1049.
(3) Walch, S. P.; Bauschlicher, C. W., Jr. In Comparison of Ab Initio
Quantum Chemistry with Experiment; Bartlett, R. J., Ed.; D. Reidel:
Dordrecht, 1985; p 17.
(4) Salahub, D. R. Adv. Chem. Phys. 1987,69, 447.
(5) Bauschlicher, C. W., Jr.; Partridge, H.; Langhoff, S. R.; Rosi, M. J.
Chem. Phys. 1991, 95, 1057.
(6) DoverstBI, M.; Lindgren, B.; Sassenberg, U.; Arrington, C. A,; Morse,
M. D. J. Chem. Phys. 1992, 97, 7087.
(7) Balasubramanian, K.; Ravimohan, Ch. J . Chem. Phys. 1990, 92,
3659.
Arrington et al.
(8) Spain, E. M.; Morse, M. D. J . Phys. Chem. 1992, 96, 2479.
(9) Morse, M. D.; Hansen, G. P.; Langridge-Smith, P. R. R.; Zheng,
L.-S.; Geusic, M. E.; Michalopoulos, D. L.; Smalley, R. E. J . Chem. Phys.
1984.80, 5400.
(10) Taylor, S.; Spain, E. M.; Morse, M. D. J. Chem. Phys. 1990, 92,
2698.
(11) Taylor, S.; Lemire, G. W.; Hamrick, Y.; Fu, Z.-W.; Morse, M. D.
J. Chem. Phys. 1988,89, 5517.
(12) Behm, J. M.; Arrington, C. A.; Morse, M. D. J . Chem. Phys. 1993,
99, 6409.
(13) Russon, L. M.; Heidecke, S. A.; Birke, M. K.; Conceicao, J.;
Armentrout, P. B.; Morse, M. D. Chem. Phys. Lett. 1992, 204, 235.
(14) Russon, L. M.; Heidecke, S. A,; Birke, M. K.; Conceicao, J.; Morse,
M. D.; Armentrout, P. B. J. Chem. Phys., in press.
(15) Taylor, S.; Spain, E. M.; Morse, M. D. J . Chem. Phys. 1990, 92,
2710.
(16) Brewer, L. Science 1968, 162, 115. Engel, N. Kem. Maanedsbl.
1949, 30, 53, 75, 97, 105, 113. Engel, N. Powder Metall. Bull. 1964, 7, 8.
Engel, N. Am. SOC.Metals, Trans. Quart. 1964, 57, 610.
(17) Fu, Z.-W.; Lemire, G. W.; Hamrick, Y.; Taylor, S.; Shui, J.-C.;
Morse, M. D. J. Chem. Phys. 1988, 88, 3524.
(18) Gupta, S. K.; Gingerich, K. A. J . Chem. Phys. 1979, 70, 5350.
(19) Moore, C. E. Nat'l Bur. Stand. (US.)Circ. No. 467 1971, Vols.
1-111.
(20) Gerstenkom,S.;Luc, P. AtlarduSpectred'AbsorptiondeIa Molecule
d'lode; CRNS, Paris, 1978. Gerstenkorn, S.; Luc, P. Rev. Phys. Appl. 1979,
14, 791.
(21) Clouthier, D. J.; Karolczak, J. Rev. Sci. Instrum. 1990, 61, 1607.
(22) Lessen, D. E.; Asher, R. L.; Brucat, P. J. Chem. Phys. Lett. 1991,
182, 412.
(23) Su, C.-X.; Hales, D. A.; Armentrout, P. B. Chem. Phys. Lett. 1993,
201, 199.
(24) Herzberg, G. Molecular Spectra andMolecular Structurel. Spectra
ofDiatomic Molecules; Van Nostrand Reinhold: New York, 1950; pp 321322.
(25) Lefebvre-Brion, H.; Field, R. W. Perturbations in the Spectra of
Diatomic Molecules; Academic Press: Orlando, 1986, p 39.
(26) Hales, D. A.; Armentrout, P. B. J . Cluster Science 1990, 1, 127.
(27) Miedema, A. R.; Gingerich, K. A. J . Phys. E 1979, 12, 2081.
(28) Van Zee, R. J.; Weltner, W., Jr. Chem. Phys. Lett. 1988,150,329.
(29) Verhaegen, G.; Smoes, S.; Drowart, J. J . Chem. Phys. 1964,40,239.
(30) Faegri, K., Jr.; Bauschlicher, C. W., Jr. Chem. Phys. 1991,153,399.
(31) Shim, I.; Gingerich, K. A. J. Chem. Phys. 1983, 78, 5693.
(32) Mandich, M. L.; Halle, L. F.; Beauchamp, J. L. J . Am. Chem. SOC.
1984,106,4403. Elkind, J. L.; Armentrout, P. B. J. Phys. Chem. 1985,89,
5626. Elkind, J. L.; Armentrout, P. B. Inorg. Chem. 1986, 25, 1078.
(33) Desclaux, J. P. Atomic Data Nuclear Data Tables 1973, 12, 31 1.
(34) Van Zee, R. J.; Weltner, W., Jr. Chem. Phys. Lett. 1984,107, 173.
(35) Bishea, G. A.; Marak, N.; Morse, M. D. J . Chem. Phys. 1991,95,
5618.
(36) Van Zee, R. J.; Weltner, W., Jr. High Temp. Sci. 1984, 17, 181.
(37) Spain, E. M.; Morse, M. D. J . Chem. Phys. 1992, 97, 4641.
(38) Cheeseman, M.; Van Zee, R. J.; Weltner, W., Jr. High Temp. Sci.
1988, 25, 143.
(39) Morse, M. D. Ado. Metal Semicond. Clusters 1993, 1, 83.
(40) Ran, Q.;Schmude, R. W., Jr.; Gingerich, K. A.; Wilhite, D. W.;
Kingcade, J. E., Jr. J . Phys. Chem. 1993, 97, 8535.